Warming of the climate of the south of Western Siberia manifests in water content reduction of lakes in the territory, particularly closed lakes. Accuracy and reliability of calculated values are of utmost importance for quality evaluation of calculated hydrological characteristics over actual data of hydrometeorological survey (in particular, water and liquid precipitation levels). The goal of this study is to evaluate the probability of water level changes (increasing, decreasing) and, further, predict spring floods on the basis of hydrometeorological survey data. Empirical and theoretical lake water level distribution functions for two observation periods (2013–2014 and 2014–2015) have been obtained, and statistical analysis of prediction accuracy of hydrological characteristics is performed. Root mean square errors for predicted hydrological characteristics are estimated using deviations between theoretical (predicted) and empirical dependencies.